Unsupervised training method and unsupervised training device for magnetic resonance parameter imaging model

Abstract

The invention discloses an unsupervised training method and an unsupervised training device for a magnetic resonance parameter imaging model. The unsupervised training method comprises the following steps: inputting acquired under-acquisition k space data into a to-be-trained magnetic resonance parameter imaging model to obtain parameter weighted image data; updating a loss function according to the acquired under-acquisition k space data and the parameter weighted image data; and updating the network parameters of the to-be-trained magnetic resonance parameter imaging model according to the updated loss function. According to the method and device, the loss function based on the parameter weighted image is introduced, calculation of the loss function can be completed by utilizing the under-collected k space data and the parameter weighted image data obtained according to the under-collected k space data, full-collected k space data does not need to be adopted, and unsupervised training of the model is achieved.

Description

technical field [0001] The invention belongs to the technical field of image reconstruction of magnetic resonance parametric imaging signals, and in particular relates to an unsupervised training method, a training device, a computer-readable storage medium and computer equipment for a magnetic resonance parametric imaging model. Background technique [0002] Quantitative Magnetic Resonance Parametric Mapping is an emerging tool for assessing and determining the fundamental biological properties of tissues. It is designed to measure absolute relaxation at magnetic resonance, thus providing comparable measurements across sites and time points. The most common way to obtain MR parametric maps is to acquire weighted images with varying imaging parameters (e.g., inversion time (TI) in T1 mapping, echo time (TE) in T2 mapping, or spin in T1ρ mapping. Time to Lock (TSL). The parametric map is then estimated by fitting these images pixel-by-pixel with the corresponding physical in...

AI Technical Summary

Problems solved by technology

In the existing deep learning MRI parametric imaging method, a fully acquired parameter map is required as a reference image, but it is often difficult to obtain full-acquired data in actual scanning, which is not conducive to network training

Images